Hybrid Quantum-Classical Neural Network Architecture Generation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing hybrid quantum-classical neural networks face challenges in efficiently generating optimal architectures that balance quantum and classical circuit distributions, leading to suboptimal performance in tasks such as image processing and natural language processing.

Innovation Solution

An apparatus and method for generating hybrid quantum-classical neural network architectures, which includes a generating device that stochastically distributes neurons between quantum and classical circuits based on setting information, and a performance comparison module that evaluates and refines the architecture to meet target performance metrics.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If neurons are distributed into quantum circuit or classical circuit based on probability, then the neural network architecture can be generated with flexibility and adaptability, but the consistency and reliability of the generated architecture may be compromised

Engineering Contradiction:

Improvearchitecture generation flexibilityVSAvoidarchitecture consistency

Core Design Contradiction:

Adaptability or versatilityVSReliability

Solution Approach 1:

The patent implements a feedback mechanism where the generated architecture is evaluated against target performance metrics, and the setting information is updated based on performance feedback to regenerate architectures until target performance is achieved, ensuring reliability while maintaining flexibility

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The patent dynamically adjusts probability parameters and other setting information based on performance evaluation results, allowing the system to adapt the distribution strategy to achieve both flexibility in generation and reliability in the final architecture

Inventive Principle:

Principle #35Parameter changes

2Productivity

If the neural network architecture is generated through stochastic distribution, then the search space can be explored efficiently, but the optimization process becomes more complex and time-consuming

Engineering Contradiction:

Improvearchitecture search efficiencyVSAvoidoptimization process complexity

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The patent segments the architecture generation process into distinct modules: neuron distribution determination, quantum circuit generation, classical circuit generation, and performance evaluation, making the complex optimization process more manageable and efficient

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary actions by pre-defining setting information including probability distributions and performance targets before the main generation process, which streamlines the optimization and reduces overall complexity

Inventive Principle:

Principle #10Preliminary action

3Manufacturing precision

If the generated architecture is evaluated against target performance metrics, then the quality of the architecture can be ensured, but the evaluation and refinement process increases the computational time

Engineering Contradiction:

Improvearchitecture performance qualityVSAvoidevaluation and refinement time

Core Design Contradiction:

Manufacturing precisionVSLoss of time

Solution Approach 1:

The patent uses feedback from performance evaluation to guide the regeneration process, allowing the system to focus computational resources on refining architectures that are closest to meeting target performance, thereby reducing overall time loss while ensuring quality

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The patent applies partial evaluation and refinement strategies, where not all aspects of performance need to be perfectly optimized, and the process can stop once target performance is sufficiently achieved, reducing unnecessary computational time

Inventive Principle:

Principle #16Partial or excessive action

Data Source

PatentUS20250045614A1Apparatus and method with hybrid quantum-classical neural network architecture generation

Publication Date: 2025.02.06 SAMSUNG ELECTRONICS CO LTD

AI summary

An apparatus with hybrid quantum-classical neural network architecture generation includes a generating device configured to generate a hybrid quantum-classical layer based neural network architecture based on setting information for generating a neural network architecture, wherein the generating device comprises a neuron distribution module configured to determine, based on the setting information, whether to distribute neurons in each layer of the neural network into a quantum circuit or a classical circuit, a quantum circuit generation module configured to generate a quantum circuit for the each layer, based on a result of distribution into the quantum circuit, and a classical circuit generation module configured to generate a classical circuit for the each layer, based on a result of distribution into the classical circuit.